Background: Although flucytosine is a key component of WHO-recommended induction treatment for HIV-associated cryptococcal meningitis, this antifungal agent is not widely available in low-income and middle-income countries due to limited production and cost. In 2018, a national flucytosine access programme was initiated in South Africa. We aimed to determine the effectiveness of flucytosine-containing induction regimens in routine care to motivate for the urgent registration of flucytosine and its inclusion in treatment guidelines. Methods: In this cross-sectional study, we compared outcomes of adults aged 18 years and older with incident laboratory-confirmed cryptococcal meningitis treated with or without flucytosine-containing regimens at 19 sentinel hospitals in South Africa. A case of cryptococcosis was defined as illness in an adult with: (1) positive cerebrospinal fluid (CSF) India ink microscopy; (2) a positive CSF cryptococcal antigen test; or (3) culture of Cryptococcus neoformans or Cryptococcus gattii from CSF or any other specimen. We excluded patients without a case report form, those with an unknown or negative HIV serology result, those with a recurrent episode, and those who did not receive antifungal treatment in hospital. We assessed cumulative in-hospital mortality at 14 days and 30 days and calculated the overall crude in-hospital case-fatality ratio. We used random-effects logistic regression to examine the association between treatment group and in-hospital mortality. Findings: From July 1, 2018, to March 31, 2020, 10 668 individuals were diagnosed with laboratory-confirmed cryptococcal meningitis, 7787 cases diagnosed at non-enhanced surveillance sites and 567 cases from eight enhanced surveillance sites with no access to flucytosine were excluded. Of 2314 adults with a first episode of cryptococcosis diagnosed at 19 facilities with access to flucytosine, 1996 had a case report form and of these, 1539 received induction antifungal treatment and were confirmed HIV-seropositive first-episode cases. Of 1539 patients who received antifungal therapy, 596 (38·7%) individuals received a flucytosine-containing regimen and 943 (61·3%) received another regimen. The median age was 36 years (IQR 32–43) and 906 (58·9%) participants were male and 633 (41·1%) were female. The crude in-hospital case-fatality ratio was 23·9% (95% CI 20·0–27·0; 143 of 596) in those treated with flucytosine-containing regimens and 37·2% (95% CI 34·0–40·0; 351 of 943) in those treated with other regimens. Patients admitted to non-academic hospitals (adjusted odds ratio [aOR] 1·95 [95% CI 1·53–2·48]; p<0·0001) and those who were antiretroviral treatment-experienced (aOR 1·30 [1·02–1·67]; p=0·033) were more likely to receive flucytosine. After adjusting for relevant confounders, flucytosine treatment was associated with a 53% reduction in mortality (aOR 0·47 [95% CI 0·35–0·64]; p<0·0001). Among survivors, the median length of hospital admission in the flucytosine group was 11 days (IQR 8–15) versus 17 days (13–21) in the comparison group (p=0·0010). Interpretation: In-hospital mortality among patients treated with a flucytosine-containing regimen was comparable to reduced mortality reported in patients receiving a flucytosine-containing regimen in a recent multicentre African clinical trial. Flucytosine-based treatment can be delivered in routine care in a middle-income country with a substantial survival benefit. Funding: National Institute for Communicable Diseases, a Division of the National Health Laboratory Service. Translation: For the Zulu translation of the abstract see Supplementary Materials section.